Input Manifolds for 4-input Makers

[marcusaddamsson]

I've got hundreds of hours in this game now, so I'm not a total noob. At this point, I've designed tons of different input and output manifolds... I can get completely balanced input manifold for a two-input maker just fine. And with stack of three splitters (and a bunch of splitters prior), I've got a pretty decent solution for a three-input recipe just fine (takes a bunch of room, but it works well): a stack of three splitters, where each of the ouputs directly feed three different three-mat makers, with a system of evenly-split inputs to each set of three-makers.

The 'bus' approach (a line of stacked-splitters directly feeding the next in line and the current maker) works kinda, sorta.. but only if you don't mind that it's really really inefficient and makes diagnosing a bank of makers really difficult, as the first guy in the line gets 100% and half of all the mats sent thru, regardless of how badly latter makers need it.

I've been toying with the 2-and-2 approach, where you put a stack of two splitters on the line between each maker, feeding two mats for two makers with each set, but then you need even more room for splitters prior (two-way splits, instead of three-way means more input belts)... and you've got the two ends as 'orphans' (in terms of load balancing).

The issue really is you only have one hammer in your tool belt: the splitter. And it will always split evenly, having absolutely no knowledge of anything other than a basic round-robin delivery paradigm. "Smart" and/or "Programmable" Splitters are useless for this application, as they only filter stuff. Personally, I *never* mix mats 'cuz if you get the wrong mats into manifold, you likely have to completely delete it to "clean" it out.

So the only way to get say... 9 or 18 makers working off an input source is to equally split the mat (either two-way or three-way) evenly. Getting four different mats to split thusly takes a ton of space, and loads of belts going everywhere...

Actually, now that I've posted this... there's a fairly easy solution for the Devs if they want it.

It's called Smart Bus Split-ger: it's got two inputs and two outputs. And you can adjust the ratio of the outputs to allow the 'bus' output to take X turns in a row in the round-robin; and the second input could be used to 'reinforce' a 'bus' belt that needs more mats inserted somewhere further down the 'bus', like if it was screws in a bank of heavy mod frames.

[Halliwax]

There's an even easier solution: just stop trying to do load balancing. Basic manifolds are perfectly fine and much easier to build. These complaints about "efficiency" are nonsense. The efficiency cost of a one-time buffer fill up drops to zero as your production keeps running. The time you lose building the balancer is worth more than that.

[Mister Fabulous]

It seems like your using "bus" to describe a manifold and "manifold" to describe balancers.

Anyway, lining up a row of splitters in front of the machines is fine. Yes, it takes time, possibly a long time, for everything to ramp up to running 100% of the time if you don't pre-fill them, but once it's there, it's just as efficient as meticulously balancing the input for all machines.

You can line up your splitters like this to save on space for the 4 input makers (manufacturers): https://imgur.com/7n5rgKg. The splitter hit box is 4 steps back from the mamufacturers.

[Colonel Sanders Lite]

Originally posted by Mister Fabulous:

It seems like your using "bus" to describe a manifold and "manifold" to describe balancers.

Yup.

Originally posted by Halliwax:

There's an even easier solution: just stop trying to do load balancing.

Also yup. Though there are a couple of edge cases I can think of where balancing is a really good idea. And I really literally just mean a couple.

Originally posted by marcusaddamsson:

Actually, now that I've posted this... there's a fairly easy solution for the Devs if they want it.

You know, I think I have a better idea. Add something similar to a pipe's valve which lets us throttle a belt to an arbitrary speed.

[DrNewcenstein]

Try a U-shaped machine layout instead of a row, where you feed them from the center. First splitter is equidistant from the center of each bank, reducing wait time for the ends. Put a Merger on the corners of each bank so they can receive from two lines.

[Halliwax]

I will say that sometimes load balancers are fun to build just for the hell of it. The mechanics of how to design them with unusual ratios can be interesting. But that doesn't make them useful, and certainly combining the trouble of building them with the already tedious task of arranging belt inputs to groups of manufacturers is not going to be so fun.

[marcusaddamsson]

https://en.wikipedia.org/wiki/Manifold_(fluid_mechanics)

https://en.wikipedia.org/wiki/Bus_(computing)

Yes, I'm mostly discussing "balanced" input manifolds for manufacturers.

I'll give y'all that I'm not using whatever terminologies have been agreed upon, but I'm not sure I'm wrong (not that I care over much, just clarifying). I got software and civil engineering in my background, so ya know...

I'm really surprised to find I'm the OCD guy in this situation... I guess I'll just hop back on the "bus" then.

[marcusaddamsson]

Originally posted by Colonel Sanders Lite:

You know, I think I have a better idea. Add something similar to a pipe's valve which lets us throttle a belt to an arbitrary speed.

Well, you can kinda do that by using v5 belts on the 'bus' outputs, and v3 for the maker outputs, but it doesn't stop the splitter from the round-robin distribution, so it only kinda works.

I'm gonna put my 2n2 SmartBus Splitter idea in suggestions once I think on it a bit more.

[HuMaNgUtAn]

I agree - my factory building really progressed when I learned to lose my obsession with the perfect split load balancers and start using manifolds. It saves so much in space and belt work, keeping my factories cleaner and less laggy. I used to build whole walled in "sandwich" layers in between my factory floors just to hide all the load balancing belts.

Now I do 95% of my load balancing via over or under-clocking particular machines, which gives you a lot more freedom than using splitters and mergers as you can specify the exact percentage. You can just split the output merger line from a row of machines as required and it uses no extra belts or space at all.

[HuMaNgUtAn]

..so using this method you can have 100% efficient perfectly smooth running factories with manifolds only and no hidden belt work, not too difficult to achieve for me building on a small scale.

here's an example shot since you liked the last one ;)

https://steamcommunity.com/sharedfiles/filedetails/?id=2561941343

the line of smelters at top left have one machine underclocked to achieve perfect numbers in 2 directions.

[marcusaddamsson]

@HuMaNgUtAn... we play very different games.. but I gotta respect the advice. Danke.

Several times now I've come to this forum for advice and/or inspiration, and often received it. It's a solo game, and prone to rabbit holes (which I'm prone to explore... sigh).

Thanks y'all for indulging me.

PS, I'm still gonna make "balanced" input manifolds for the first stage makers, like refineries and foundries and such. But in general I get what y'all are sayin'...

[Colonel Sanders Lite]

Originally posted by marcusaddamsson:

Originally posted by Colonel Sanders Lite:

You know, I think I have a better idea. Add something similar to a pipe's valve which lets us throttle a belt to an arbitrary speed.

Well, you can kinda do that by using v5 belts on the 'bus' outputs, and v3 for the maker outputs, but it doesn't stop the splitter from the round-robin distribution, so it only kinda works.

I'm gonna put my 2n2 SmartBus Splitter idea in suggestions once I think on it a bit more.

I'm already doing that sort of thing and it helps but it really only solves the basic issue in the rare situation where a machine draws exactly 60, 120, 270, or 480 mats/min.

As an arbitrary example, let's suppose you have an array of machines each drawing 12.34 ingots/min. If you could set a governor on each feed belt which would limit the feed rate to 12.34 ingots/min to match the machine, all machines would then be getting exactly what they need and the ingot buffers would be unable to fill up until the outputs where backed up.

Originally posted by marcusaddamsson:

I'll give y'all that I'm not using whatever terminologies have been agreed upon, but I'm not sure I'm wrong (not that I care over much, just clarifying).

You're not *technically* wrong but there are reasons for distinguishing these specific terminologies as they apply to this style of game. For example - While a balanced manifold is technically still a kind of manifold, balancing is not actually needed *most* of the time so it's important to distinguish the rare niche situation where it is important that the belts are balanced. Because of this, the two terms evoke very different requirements.